ABSTRACT The Matlab Linked Database (MLD) will provide the population health sciences with an unprecedented opportunity to understand the relationships between health and human welfare over time and across generations. Matlab, Bangladesh, is one of the few settings in the world that combines well-documented quasi- randomization of interventions, long duration of follow-up (40 years), and detailed tracking of attrition across an unusually rich set of outcomes. A population of over 200,000 people in 141 villages has been followed since the mid-1970s by the International Centre for Diarrheal Disease Research, Bangladesh (icddr,b) through the Matlab Health and Demographic Surveillance System (MHDSS). Matlab is also the setting of multiple high- impact interventions, including the highly successful Matlab Maternal Health and Family Planning (MCH/FP) program. In 1996, the widely-utilized Matlab Health and Socioeconomic Survey (MHSS1) was conducted on a representative sample of 7% of households in the Matlab study area, with P01 funding from NIA and NICHD. In 2012-4, with NIA R01 funding, MHSS2 followed the MHSS1 primary sample respondents and their descendants, focusing on the health and well-being of children born during the MCH/FP intervention period as they entered adulthood, and their mothers who were entering old-age. Among many other advantages, MHSS2 conducted unusually successful migrant tracking, capturing 92-94% of surviving members of all key age-sex cohorts. MHSS1 and MHSS2 included community/facility surveys for the entire MHDSS area. While MHSS1/2 are valuable in their own right, their impact is amplified by linkage to an array of longitudinal, data collected over 40 years through MHDSS. MLD would embed MHSS1/2 data within decades of socioeconomic census, demographic, kinship and programmatic uptake data from the MHDSS. Specifically, icddr,b has agreed to contribute data not just on individuals in the MHSS1/MHSS2 samples, but on a larger sampling frame including individuals from as far back as 1974 who out-migrated or died before MHSS1 or between survey rounds. Such data will enable a wide range of studies on the role of selection, attrition, household formation and weighting that will have impacts for evaluation studies globally. The proposed archiving study has four specific aims: 1) To release and support a well-documented Matlab Linked Database (MLD) for public and restricted access; 2) to securely link MHSS1 and MHSS2 and create harmonized MHSS1/2 data; 3) to embed MHSS1/2 data in a larger 40-year sampling frame that will include census, vital registration; kinship linkages and other MHDSS data for surveyed and unsurveyed individuals; and 4) to release a comprehensive and user-friendly exposure files on facilities, programs and services that will facilitate analysis of multiple interventions, mediators and confounders occurring in the study site. Taken together, MLD will allow users to understand the long-term consequences of health and development interventions on a wide range of outcomes, across multiple generations along a 40-year time horizon.